[FFmpeg-devel] [PATCH 6/6] jrev/xvid: hardcode use of C put/add_pixels_clamped.

[Ronald S. Bultje]

This removes the last use of the ff_put/add_pixels_clamped global
function pointers, and as such they are removed.

This patch has a negative effect on performance on MIPS, since there's
a SIMD-optimized put/add_pixels_clamped, but no xvid or jrev. From a
code maintenance point of view, that is probably acceptable.

Because the global function pointers are removed, this fixes the following
tsan warnings when running e.g. fate-dnxhr-parse:

WARNING: ThreadSanitizer: data race (pid=29917)
  Write of size 8 at 0x0000025b12d8 by thread T2 (mutexes: write M1543):
    #0 ff_idctdsp_init src/libavcodec/idctdsp.c:313 (ffmpeg+0x00000044b68e)
[..]
  Previous write of size 8 at 0x0000025b12d8 by thread T1 (mutexes: write M1541):
    #0 ff_idctdsp_init src/libavcodec/idctdsp.c:313 (ffmpeg+0x00000044b68e)
---
 libavcodec/idctdsp.c  | 18 ++++++------------
 libavcodec/idctdsp.h  |  6 ++++--
 libavcodec/jrevdct.c  |  4 ++--
 libavcodec/xvididct.c |  4 ++--
 4 files changed, 14 insertions(+), 18 deletions(-)

diff --git a/libavcodec/idctdsp.c b/libavcodec/idctdsp.c
index 84dd645..d596aed 100644
--- a/libavcodec/idctdsp.c
+++ b/libavcodec/idctdsp.c
@@ -80,11 +80,8 @@ av_cold void ff_init_scantable_permutation(uint8_t *idct_permutation,
     }
 }
 
-void (*ff_put_pixels_clamped)(const int16_t *block, uint8_t *pixels, ptrdiff_t line_size);
-void (*ff_add_pixels_clamped)(const int16_t *block, uint8_t *pixels, ptrdiff_t line_size);
-
-static void put_pixels_clamped_c(const int16_t *block, uint8_t *av_restrict pixels,
-                                 ptrdiff_t line_size)
+void ff_put_pixels_clamped_c(const int16_t *block, uint8_t *av_restrict pixels,
+                             ptrdiff_t line_size)
 {
     int i;
 
@@ -157,8 +154,8 @@ static void put_signed_pixels_clamped_c(const int16_t *block,
     }
 }
 
-static void add_pixels_clamped_c(const int16_t *block, uint8_t *av_restrict pixels,
-                                 ptrdiff_t line_size)
+void ff_add_pixels_clamped_c(const int16_t *block, uint8_t *av_restrict pixels,
+                             ptrdiff_t line_size)
 {
     int i;
 
@@ -290,9 +287,9 @@ av_cold void ff_idctdsp_init(IDCTDSPContext *c, AVCodecContext *avctx)
         }
     }
 
-    c->put_pixels_clamped        = put_pixels_clamped_c;
+    c->put_pixels_clamped        = ff_put_pixels_clamped_c;
     c->put_signed_pixels_clamped = put_signed_pixels_clamped_c;
-    c->add_pixels_clamped        = add_pixels_clamped_c;
+    c->add_pixels_clamped        = ff_add_pixels_clamped_c;
 
     if (CONFIG_MPEG4_DECODER && avctx->idct_algo == FF_IDCT_XVID)
         ff_xvid_idct_init(c, avctx);
@@ -310,9 +307,6 @@ av_cold void ff_idctdsp_init(IDCTDSPContext *c, AVCodecContext *avctx)
     if (ARCH_MIPS)
         ff_idctdsp_init_mips(c, avctx, high_bit_depth);
 
-    ff_put_pixels_clamped = c->put_pixels_clamped;
-    ff_add_pixels_clamped = c->add_pixels_clamped;
-
     ff_init_scantable_permutation(c->idct_permutation,
                                   c->perm_type);
 }
diff --git a/libavcodec/idctdsp.h b/libavcodec/idctdsp.h
index f9ba6c3..26221f6 100644
--- a/libavcodec/idctdsp.h
+++ b/libavcodec/idctdsp.h
@@ -97,8 +97,10 @@ typedef struct IDCTDSPContext {
     enum idct_permutation_type perm_type;
 } IDCTDSPContext;
 
-extern void (*ff_put_pixels_clamped)(const int16_t *block, uint8_t *pixels, ptrdiff_t line_size);
-extern void (*ff_add_pixels_clamped)(const int16_t *block, uint8_t *pixels, ptrdiff_t line_size);
+void ff_put_pixels_clamped_c(const int16_t *block, uint8_t *av_restrict pixels,
+                             ptrdiff_t line_size);
+void ff_add_pixels_clamped_c(const int16_t *block, uint8_t *av_restrict pixels,
+                             ptrdiff_t line_size);
 
 void ff_idctdsp_init(IDCTDSPContext *c, AVCodecContext *avctx);
 
diff --git a/libavcodec/jrevdct.c b/libavcodec/jrevdct.c
index 89dd9f2..3b15a52 100644
--- a/libavcodec/jrevdct.c
+++ b/libavcodec/jrevdct.c
@@ -1159,11 +1159,11 @@ void ff_j_rev_dct1(DCTBLOCK data){
 void ff_jref_idct_put(uint8_t *dest, ptrdiff_t line_size, int16_t *block)
 {
     ff_j_rev_dct(block);
-    ff_put_pixels_clamped(block, dest, line_size);
+    ff_put_pixels_clamped_c(block, dest, line_size);
 }
 
 void ff_jref_idct_add(uint8_t *dest, ptrdiff_t line_size, int16_t *block)
 {
     ff_j_rev_dct(block);
-    ff_add_pixels_clamped(block, dest, line_size);
+    ff_add_pixels_clamped_c(block, dest, line_size);
 }
diff --git a/libavcodec/xvididct.c b/libavcodec/xvididct.c
index 4642a30..d8f3dd7 100644
--- a/libavcodec/xvididct.c
+++ b/libavcodec/xvididct.c
@@ -321,13 +321,13 @@ void ff_xvid_idct(int16_t *const in)
 static void xvid_idct_put(uint8_t *dest, ptrdiff_t line_size, int16_t *block)
 {
     ff_xvid_idct(block);
-    ff_put_pixels_clamped(block, dest, line_size);
+    ff_put_pixels_clamped_c(block, dest, line_size);
 }
 
 static void xvid_idct_add(uint8_t *dest, ptrdiff_t line_size, int16_t *block)
 {
     ff_xvid_idct(block);
-    ff_add_pixels_clamped(block, dest, line_size);
+    ff_add_pixels_clamped_c(block, dest, line_size);
 }
 
 av_cold void ff_xvid_idct_init(IDCTDSPContext *c, AVCodecContext *avctx)
-- 
2.8.1